By Howard A. Roberson
Fig. 1--Typical frequency response curves of a cassette deck with and without (---) Dolby noise reduction, showing effects of three different bias settings.
Whenever Audio reports on the results for testing a tape deck, we include a number of figures to help ex plain what the recorder's capabilities are. We usually have three graphs, or plots, that show the frequency response and headroom for three different tape types, usually a Type I (ferric), a Type II (CrO2 type), and a Type IV (metal). Sometimes, a Type III (FeCr) is used in addition to, or in place of, one of the above.
Strictly speaking, what is commonly called "frequency response" would be more correctly labeled "amplitude response versus frequency." We are concerned with, and what our figures do show, is how the level, or amplitude, of the playback varies across the frequency band. Other high-fidelity components are rated by their frequency responses, of course, but a recorder's response is a long way from the ex tended, ruler-flat response of an amplifier. One of the goals of this article is to provide some guidelines for examining such graphs with greater perception.
It is best to compare the performance of one deck to another with the same signal or flux-level reference, and to use a reference easily identified by the user or reader. The Dolby standard flux level (200 nWb/m at 400 Hz) is used as the reference level of frequency response, as well as for other tests. Substantially all cassette decks have those little double-D symbols which pinpoint the record level that should gain Dolby flux levels. (It should be noted that because of equalization effects, the 200 nWb/m flux-level figure is correct for 400 Hz only.) When frequency responses are run at the reference level, there will be a high-end dropping caused by any combination of several things: The tape itself, the design of the record head, the amount of equalization used in record--well, there's saturation and self-erasure mixed in here too. In any event, the plots at Dolby level are important since they show what the high-level, high-frequency limits are for the recorder/tape combination.
Responses are also shown at a level 20 dB lower, where frequency responses are usually specified by makers. This does not mean, of course, that with the record level 20 dB below Dolby level, we will be certain to get that ruler-flat line. There will be both lower and upper frequency limits, with electrical, mechanical and magnetic factors all involved. The results at this level are affected, perhaps greatly, by the closeness of match between the bias current provided by the machine and the bias needs of the tape formulation. The performance in Dolby mode is affected by bias discrepancies, as well as any record sensitivity offsets.
Because most users operate their decks in Dolby mode the majority of the time, the responses with NR are considered the normal ones. Responses without Dolby are usually also shown, particularly when there is a noticeable difference.
Examination of the figure with this article reveals that there is a plot at Dolby level and three plots at -20 dB, these last with three different bias conditions. The test signal was a sine wave which was swept from 20 Hz to 20 kHz. By this means, each and every point is a true data point. Discrete-point data are required if it is necessary to go beyond these limits in the usual frequency response curve.
Let's start with a critical look at the Dolby-level response. Our reference frequency for plotting is 1 kHz, and the trace crosses the reference-level line there. There is a slight rise as we go down to 100 Hz, below which there are some bumps from head-contour effects. The response falls below 40 Hz, but it is not down 3 dB until just above 20 Hz. It would be ideal if there were no head-contour effects, and it is desirable that deviations from 40 Hz up be less than ± 1.5 dB. The significance of the lower limit is twofold:
How low in frequency do your music sources go, and are the levels high in this region? Sometimes, but not in the case shown, recording at Dolby level causes a further dropping of the low end, relative to the -20 dB results. This is indicative of low-frequency over load or saturation and shows there is a definite level limit.
It's best if the response continues flat above 1 kHz, but at some point there will be a headroom limitation, and the response will roll off. In the figure, response is down 3 dB at about 6.3 kHz, which would be limiting in any attempt to record at high levels any music with considerable high-frequency content. The distortion also rises rapidly with level in the roll-off region, so it's not just a case of losing a little signal level. We would certainly like to see the response quite flat from 100 Hz to close to 10 kHz, which is possible with many decks. A general tilt, such as is evident from 70 Hz to say 4 kHz, might make the sound a little extra bassy.
Now let's take a look at the group of three sets of responses at -20 dB. The labels show what the deck's relative bias levels would be for a particular tape to get the results plotted. In each case, the trace without Dolby NR is shown dashed. We can see that for each bias setting, the plot with NR is not quite as flat as without, and that when the bias matches the tape, the deviations are at a minimum. (The plot at Dolby level includes traces in both modes; the responses were exactly the same.) At-20 dB the deck is not operating in a condition of high-frequency over load, so there cannot be a basis for excusing a general roll-off, such as shown in the high-bias case. At the same time, excessive high-end response can be much too bright, with a higher distortion level. So, tape formulations that match the deck are selected for testing. Immediately, we can say that we would like to see a response that is flat within a dB or so, from at least 100 Hz to 10 kHz, and extensions down to 40 Hz or so and up to 13 kHz or more would be nice to have. A very important region is from about 700 Hz to about 6 kHz where Dolby mistracking can cause either excessive presence, because of a boost in this region, or a dull, remote sound because of a saddle-like droop.
In summary, look at the responses at both Dolby level and at-20 dB over the range from about 40 Hz to 10 kHz or so. In both cases, flatness is the goal. Also give attention to head bumps at the low end, headroom limitations at the higher level, and Dolby-mode boost or saddle deviations at the lower level. See how far down in frequency the responses actually extend and whether there is additional roll-off at Dolby level. The extension of flat response above 13 kHz at -20 dB has some value, but this is not as significant as the other times, in most cases.
(Source: Audio magazine, Jan. 1981)
Also see:
Performance of High Energy in Magnetic Materials in Audio Cassette Recording Tapes (Sept. 1978)
= = = =